I was interviewed by a really cool program called ‘Pint of Science’ that is spearheaded by some really incredible graduate students and postdocs at Moffitt. They take a scientist to a pub, buy them a beer, and get them to talk about their research in terms that the general public will be able to follow. They are doing a really great job with outreach. We need more of this!
Scientific impact vs populat impact of 50 famous scientists.
One issue is that citation indeces are not that useful to compare scientists across disciplines. In any case I would be interested to see plots like this with a larger cohort of scientists whithin a given field of science.
Analogies are powerful and dangerous. It is a good idea to not take more of it than what the analogy can offer. But the idea that building mathematical models can be compared to map making is a powerful one. If you look at the image featured in this post it is easy to realise that maps, like mathematical models, capture the assumptions and understanding of a community…even if they are wrong or incomplete.
Maps show us our biases. The image above (top left) shows the map using a conventional representation. It uses the Mercator projection where Greenland can be seen as a much bigger landmass than Australia. Which is not true. To the right you will seethe Gall-Peters projection which does a better job but is not perfect either. In general all these maps are wrong in one respect or the other but nobody should doubt that they are useful.
Here is another map:
It is a map of the subway system in London and as an example it captures one important aspect of mathematical modelling and map making: the important thing about a map like this is not only what it shows but what it chooses not to. This map is incredibly useful and a good part of that is because it shows only the information that a traveller needs in order to reach any corner of London served by the tube. Arguably it could not show less information (maybe the thick blue line showing the Thames river) but even more importantly: adding more information would likely make it a worse map. Some people might argue that there is a lot of information that is critical if we want to understand transportation in London. Maybe the nature of the terrain that had to be open for building the lines? The elevation? how congested those lines typically are? distance between stops (the map is not to scale), average travelling times? They all sound like good ideas but they would make the map more complicated and more difficult to navigate.
A couple of years ago a few of us at Moffitt took part in what could be described as a series of synchronised and competitive hackathons. Basically we were put into different groups and asked to come with a question and a mathematical model that could be used to personalise treatments in the clinic. My group? We went on to see how mathematical models could be used to treat prostate cancer patients (2nd most common cancer among males) with metastases in the bone (which happens in 90% of the cases of patients that die of the disease).
We recently published our ideas in a paper in Clinical and Experimental Metastases (online here and soon in bioarXiv). There we describe a mathematical model that captures the interactions between a heterogeneous metastasis (in terms of three possible mutations) and the bone microenvironment (see the figure above). The features of the tumour cells can be parameterised with patient-specific information whereas the rest of the parameters of this model were taken from literature.
What can you do with a model like this? I guess a lot of things but what we did is the following. We created a genetic algorithm, a tool that can optimise treatments, and we used it to see if we could optimise a sequent of treatments (including both conventional as well as new targeted treatments) for specific prostate cancer patients with metastases to the bone.
We are currently working on the validation of this approach using retrospective data but I am happy to see that the ideas are out there for any other team of scientists to use them as well. Our team included pathologists, oncologists, radiotherapy specialists, surgeons, experimental biologists and, of course, mathematical and computational biologists. It does take a diversity of talents to make these type of approaches work but it is worth the effort.
Will this be better than the current approach used in the clinics and cancer hospitals? Our current results suggest that this is possible. Moreover, we think that mathematical models (that captures the biology of the disease better than current standards) can be the glue that joins biological understanding with clinical data and that alone should help move medicine from craft to science.
More than an apology a defense of theory and “theorists”.
After a long hiatus, Artem comes back with a post where he refocuses his scientific interests. A theorist, a gadfly that questions our assumptions and makes us aware of our blind spots. Now, gadflies might not be pleasant but, in science, they are a necessary good.
Almost four months have snuck by in silence, a drastic change from the weekly updates earlier in the year. However, dear reader, I have not abandoned TheEGG; I have just fallen off the metaphorical horse and it has taken some time to get back on my feet. While I was in the mud, I thought about what it is that I do and how to label it. I decided the best label is “theorist”, not a critical theorist, nor theoretical cognitive scientist, nor theoretical biologist, not even a theoretical computer scientist. Just a theorist. No domain necessary.
The problem with a non-standard label is that it requires justification, hence this post. I want to use the next two thousand words to return to writing and help unify my vision for TheEGG. In the process, I will comment on the relevance of philosophy to science, and the theorist’s integration of…